PREDICTING THE SHEAR CAPACITY OF REINFORCED CONCRETE SLENDER BEAMS WITHOUT STIRRUPS BY APPLYING ARTIFICIAL INTELLIGENCE ALGORITHMS

Abstract

This paper focusses on the ongoing discussion of developing a single relationship that can accurately predict the shear capacity of slender, reinforced concrete (RC) beams without stirrups. To date, the main approach used to predict the shear capacity of RC beams, has been based on the derivation of a formula from experimental data. In this study, the approach uses the development of RC FEM models without stirrups, where the beam width is larger or equal to the section height and tested under three-point bending. The models were created and analysed by using Reconan FEA software, where the obtained results from the nonlinear analyses were used to construct a large database of 10,000 beams with varying material and geometric properties. Artificial Intelligence (AI) training was performed by using machine learning algorithms on the numerically generated database to develop predictive models and to develop an improved formula for predicting the shear capacity of RC beams without stirrups. The proposed predictive formula was validated against an available ACI database of RC beams that were assembled by using experimentally tested, physical beams without stirrups. The predictive formula was also compared with the design code formulae proposed by ACI 318-19 and Eurocode 2. According to the numerical findings of this research work, the proposed formula outperformed both design formulae demonstrating significant potential in replacing the current design approach.